Deep learning: A dense layer

Compute a whole layer of neurons at once.

What to build

A program that applies a fully connected layer to several input samples.

Requirements

  • The layer has m neurons and n inputs: W is m × n, b has m values, and the activation is relu, sigmoid, tanh or linear.
  • For each sample, print the m outputs with four decimal places, separated by spaces.

Input

Line 1: m, n and the activation. Then m rows of W. Then the biases. Then k and k samples.

Output

k lines.

Constraints

m, n ≤ 20; k ≤ 50.

Example

Input:

2 2 relu
1 -1
0.5 0.5
0 1
2
2 1
0 3

Output:

1.0000 2.5000
0.0000 2.5000

Why it works

Every neuron reads the same inputs with its own weights, so a layer is a matrix-vector product followed by the activation.

Examples

Example 1 · Input
2 2 relu
1 -1
0.5 0.5
0 1
2
2 1
0 3
Expected output
1.0000 2.5000
0.0000 2.5000
Example 2 · Input
1 1 linear
2
-1
3
0
1
-2
Expected output
-1.0000
1.0000
-5.0000

Prompts inside input() are ignored when checking, so input("n: ") is fine.

Hints

Hint 1

Each output is one row of W dotted with the input, plus that row's bias, then the activation.

Hint 2

Use the same function for every sample.

Solution

Show one correct solution (try it yourself first)
import math

m, n, activation = input().split()
m = int(m)
W = [list(map(float, input().split())) for _ in range(m)]
b = list(map(float, input().split()))
f = {
    "relu": lambda v: max(0.0, v),
    "sigmoid": lambda v: 1 / (1 + math.exp(-v)),
    "tanh": math.tanh,
    "linear": lambda v: v,
}[activation]
for _ in range(int(input())):
    x = list(map(float, input().split()))
    out = [f(sum(w * xi for w, xi in zip(row, x)) + bi) for row, bi in zip(W, b)]
    print(" ".join(f"{v:.4f}" for v in out))

Lesson: Neurons, layers and activation functions →